ESP32-Based Inventory Tracker for Goods Recording with Barcode Identification

Sitti Wetenriajeng Sidehabi (Politeknik ATI Makassar)
Wahidah Wahidah (Politeknik ATI Makassar)
St. Nurhayati Jabir (Politeknik ATI Makassar)
Muh Rizal Wahyudin (Politeknik ATI Makassar)
Muhammad Fauzan Suharman (Politeknik ATI Makassar)

Research in Education, Technology, and Multiculture
Vol 5, No 3 (2026): Research in Education, Technology, and Multiculture

Manual inventory recording remains vulnerable to delayed updates, transcription errors, and repeated reconciliation, particularly in warehouse units that still rely on spreadsheet-based data entry. This study aims to design and evaluate a Smart Inventory Tracker for recording goods at the PT Berkah Industri Mesin Angkat Site Kendari by integrating barcode scanning, ESP32-based edge processing, and cloud-based storage. The prototype used an ESP32 development board, a GM66 barcode scanner communicating through a Universal Asynchronous Receiver-Transmitter (UART) interface, a thin-film transistor (TFT) display driven through the Serial Peripheral Interface (SPI) protocol, a 12 V DC power supply with step-down regulation, a JSON reference database hosted on GitHub, and a Google Apps Script endpoint for writing validated transactions to Google Spreadsheets. The evaluation followed an engineering research approach covering hardware assembly, software integration, database matching, transaction logging, and functional testing. System performance was assessed through barcode readability distance testing, registered and unregistered barcode validation, end-to-end response-time measurement, and paired comparison with manual Google Sheets recording. Barcode reading was reliable at 3.7-16.4 cm, while scans at 3.6 cm and 16.5 cm failed outside the empirically verified optical working range. Ten registered barcode samples were successfully displayed and recorded, with an average end-to-end cloud logging time of 4.58 s per item. Manual recording required 45.57 s per item on average; therefore, the automated method reduced the observed recording duration by 40.99 s per item, equivalent to an 89.94% time reduction. Unregistered barcodes were rejected and were not appended to the spreadsheet. These findings indicate that the architecture supports faster and more consistent recording for small- to medium-sized warehouse operations. However, component-level latency logging, offline buffering, and broader network-condition testing remain necessary before large-scale commercial deployment. Keywords: Inventory management, ESP32, Barcode identification, Internet of Things (IoT), Google spreadsheets.